Skip to content
IAM Insights

Explainers · Glossary

Workload Identity Certificate

WIC

The WIMSE draft’s X.509 credential carrying one URI subject alternative name that identifies a workload within its trust domain.

reference →

In our briefings

The claim each of the newest briefings on Workload Identity Certificate would have you leave with.

  1. Schrock's draft gives agent-action records one way to carry a human's approval. By reference, the record holds a SHA-256 digest of the approval artifact's canonical bytes; embedded, it carries a claim with each named approver's own signature and the action digest. A record field that merely asserts "a human approved" MUST NOT be treated as evidence, and an approval for a different action MUST invalidate the binding, not merely weaken it. When the embedded form asks for a quorum, it counts distinct accountable humans, not distinct signatures.

    A 'human approved' field is not evidence, says Schrock's draft; only a bound, signed approval countsBI-FIND-2026-1882

  2. Under revision -03, a WIMSE credential that validates against its trust domain's anchors proves membership, not identity. Section 9.2, new in -03, says that check "establishes that the peer belongs to that trust domain, but does not by itself establish which workload the peer represents", for clients and servers alike, and adds: "Proof of possession does not mitigate this vulnerability." Telling workloads apart takes the identifier comparison that section 1.2 now spells out. A caller checks the server certificate against the identifier it expects, as RFC 9525 describes, and a callee checks the audience of the caller's proof. Section 5.2 says the callee MUST make that audience check against trusted configuration "rather than request fields such as Host", because intermediaries may rewrite the request URI. Where several instances share one identifier, the identifier alone cannot tell them apart for authorization or auditing.

    WIMSE credentials -03: a valid signature proves trust-domain membership, not which workload is callingBI-FIND-2026-1907

  3. Reading the public documentation for AWS Nitro Enclaves, Google Confidential Space and Azure's attestation service, Reddy, Ritz, Sheffer and Mihalcea find that the measurements each provider attests vouch for an approved workload and cannot tell two running copies of it apart. In their words, "Measurements establish the type of a workload, not which running instance it is." That suffices to release a secret to any genuine instance of an approved image, but attested TLS may need to reach one specific instance. Where copies must be told apart, the draft has the workload generate a key pair inside the attested environment and carry the public key in the attestation, as Nitro already lets an enclave do in its attestation document. The authors leave the SEAT working group to decide whether attested TLS must identify a specific instance, and whether a key-based identity is enough.

    Attestation shows which image is running, not which copy; a new IETF draft weighs two ways to tell copies apartBI-FIND-2026-1731

Cited in 5 briefings

Every briefing that discusses this term, newest first.

  1. A 'human approved' field is not evidence, says Schrock's draft; only a bound, signed approval counts

    Draft · · IETF Datatracker · Iman Schrock, EMILIA Protocol · Jul 3, 2026

  2. WIMSE credentials -03: a valid signature proves trust-domain membership, not which workload is calling

    Draft · · IETF Datatracker · Brian Campbell et al. · Jul 2, 2026

  3. Attestation shows which image is running, not which copy; a new IETF draft weighs two ways to tell copies apart

    Draft · · IETF Datatracker · Tirumaleswar Reddy.K et al. · Oct 28, 2026

  4. Six companies' agent-auth draft, since adopted by the WIMSE group, keeps credentials out of the model's reach

    Draft · · IETF Internet-Draft (individual submission, since adopted by the WIMSE working group) · Pieter Kasselman et al. · Jul 6, 2026

  5. The IETF's simplest workload proof proves the key and leaves the request unprotected, and the draft says so itself

    Draft · · IETF Datatracker · Brian Campbell et al. · Aug 27, 2026